According to various embodiments, a radar system is described comprising a radar receiver configured to receive radio signals, a range Fourier transform stage configured to generate, for each of a plurality of chirps, a vector of range Fourier transform coefficients, a machine learning model configured to generate, for each of one or more additional chirps, a vector of range Fourier transform coefficients from the vectors of Fourier transform coefficients generated for the plurality of chirps, and an object detector configured to perform velocity estimation of one or more detected objects using the Fourier transform coefficients generated by the machine learning model.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A radar system, comprising: A radar receiver configured to receive radio signals, A range Fourier transform stage configured to generate, for each of a plurality of chirps, a vector of range Fourier transform coefficients, A machine learning model configured to generate, for each of one or more additional chirps, a vector of range Fourier transform coefficients from the vectors of Fourier transform coefficients generated for the plurality of chirps, and An object detector configured to perform velocity estimation of one or more detected objects using the Fourier transform coefficients generated by the machine learning model.
2. The radar system of claim 1, wherein the object detector comprises a Doppler Fourier transform stage configured to generate Doppler Fourier transform coefficients from the vectors of range Fourier transform coefficients for the plurality of chirps and for the one or more additional chirps and wherein the object detector is configured to perform the velocity estimation using the Doppler Fourier transform coefficients.
3. The radar system of claim 1, wherein each chirp of the plurality of chirps and the one or more additional chirps is associated with a respective chirp time of a sequence of chirp times and wherein the machine learning model is configured to generate, for each additional chirp, the vector of range Fourier transform coefficient for the chirp as if the chirp were sent at the chirp time associated with the additional chirp.
4. The radar system of claim 3, wherein the radar system is configured to transmit each chirp of the plurality of chirps at the chirp time associated with the chirp.
5. The radar system of claim 3, wherein the sequence of chirp times is a uniform sequence of chirp times.
6. The radar system of claim 3, wherein at least some of the additional chirps are associated with chirp times between chirp times associated with chirps of the plurality of chirps.
7. The radar system of claim 3, wherein at least some of the additional chirps are associated with chirp times before or after the chirp times associated with the chirps of the plurality of chirps.
8. The radar system of claim 1, wherein the machine learning model is configured to generate, for each of the one or more additional chirps, the vector of range Fourier transform coefficients from input data comprising the vectors of Fourier transform coefficients generated for the plurality of chirps, wherein the input data comprises zero entries for the range Fourier transform coefficients for the one or more additional chirps and the machine learning model is configured to fill the zero entries with range Fourier transform coefficients for the one or more additional chirps.
9. The radar system of claim 1, wherein the machine learning model is a machine learning model configured to update the range Fourier transform coefficients for the plurality of chirps and the object detector is configured to perform velocity estimation of the one or more detected objects using the updated range Fourier transform coefficients for the plurality of chirps.
10. The radar system of claim 1, comprising a training device configured to train the machine learning model.
11. The radar system of claim 10, wherein the training device is configured to train the machine learning model by supervised learning wherein the training device is configured to generate ground truth data having a multiplicity of ground truth data elements by simulation of received radio signals and processing the multiplicity of ground truth data to a range FFT output or simulating a range FFT output.
12. The radar system of claim 11, wherein each ground truth data element comprises, for each of the one or more additional chirps, a ground truth vector of range Fourier transform coefficients.
13. The radar system of claim 12, wherein each ground truth data element additionally comprises, for each of the one or more chirps, a ground truth vector of range Fourier transform coefficients.
14. The radar system of claim 11, wherein the training device is configured to generate the multiplicity of ground truth data elements by varying at least one of a number of targets of a scene for which the ground truth data elements are generated, locations of the respective number of targets in the scene, radar cross sections of the targets in the scene and random errors and noise and to train the machine learning model using the multiplicity of ground truth data elements.
15. The radar system of claim 1, wherein the machine learning model is a neural network.
16. The radar system of claim 15, wherein the machine learning model comprises at least one of a multi-layer perceptron, an autoencoder and a recurrent neural network.
17. A method for performing velocity estimation in a radar system, the method comprising: Receiving radio signals; Generating, for each of a plurality of chirps, a vector of range Fourier transform coefficients; Generating, for each of one or more additional chirps, a vector of range Fourier transform coefficients from the vectors of Fourier transform coefficients generated for the plurality of chirps; and Performing velocity estimation of one or more detected objects using the Fourier transform coefficients generated by the machine learning model.
18. The method of claim 17, further comprising training the machine learning model.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 27, 2022
February 18, 2025
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.